Based on a trace of movements of a movable object such as a vehicle, a navigation apparatus learns a new road which is not registered in a ready-made map data. Such a navigation apparatus is conventionally disclosed in, for example, Japanese Patent Laid-Open No. H06-88733 (JP '733) and Japanese Patent Laid-Open No. 2006-125883 (JP '883). The road learn function is a function that learns a new road not registered in the map data, by generating a new road (i.e., a learned road) and storing it, based on the trace of the vehicle's movements from a point where a current vehicle position departed from an existing road of the map data to a point where the current vehicle position returned to the existing road.
In the prior art, the learned road links, which are based on the vehicle's travel experience, and the existing road links in the ready-made map data are evenly reflected, without distinction, to the contents of the route guidance, when the route guidance is provided for the user. However, when the learned roads are reflected in the route guidance, such learned roads may not necessarily be appropriate as a route guidance object, which is caused by the lack of reliability, or uncertainty, of the learned roads.
For instance, the data stored as the learned road link regarding the shape and/or the progress direction of such road may be different from the actual shape and/or direction due to its generation method nature, which is based on the vehicle's travel experience, or the trace of the vehicle movements. Therefore, in such case, the route guidance provided to the user may be different from what the user is actually experiencing, in terms of, for example, the provided shape of the branch point, thereby confusing the user rather than guiding.